INTRODUCTION
An ecosystem approach to fisheries management requires not only evaluation of fish stocks but also assessment of the state of ecosystems in order to monitor the integrity of the marine habitats influenced by fishing operations (FAO, 2003). Thus, practical methods are needed to evaluate and monitor the overall health of the ecosystem (FAO, 2003; Garcia et al., Reference Garcia, Zerbi, Aliaume, Do Chi and Lasserre2003). Benthic invertebrates are considered to be suitable organisms for assessing ecosystem quality and environmental changes (Gray & Mirza, Reference Gray and Mirza1979; Gray et al., Reference Gray, Clarke, Warwick and Hobbs1990, Reference Gray, McIntyre and Stirn1992; Rosenberg & Resh, Reference Rosenberg and Resh1993) because they occur in a wide variety of forms and habitats and are known to respond to many types of environmental pressures, e.g. pollution, habitat degradation and disturbance (Pearson & Rosenberg, Reference Pearson and Rosenberg1978; Mirza & Gray, Reference Mirza and Gray1981; Warwick et al., Reference Warwick, Pearson and Rushwahyuni1987; Gray et al., Reference Gray, McIntyre and Stirn1992, Reference Gray, Dayton, Thrush and Kaiser2006). Benthic invertebrates tend to be confined to relatively small areas due to their limited mobility and effectively integrate historical environmental conditions (Warwick, Reference Warwick1993; Salas et al., Reference Salas, Marcos, Neto, Patricio, Pérez-Ruzafa and Marques2006). For example, certain species can suffer from the effects of predators or competitors or their own inability to tolerate the prevailing physical conditions. Only species having traits optimized for the altered environment will survive (Townsend & Hildrew, Reference Townsend and Hildrew1994). Certain combinations of general biological traits and life history strategies are favoured, depending on the environmental characteristics of the habitats in question (Southwood, Reference Southwood1977). Understanding the link between species traits and environmental variability will help to predict community responses (Bremner et al., Reference Bremner, Rogers and Frid2006a, Reference Bremner, Rogers and Fridb; Bremner, Reference Bremner2008).
Demersal trawl fishing activities are considered to be one of the greatest sources of anthropogenic disturbance to marine benthic communities (Watling & Norse, Reference Watling and Norse1998; Kaiser et al., Reference Kaiser, Ramsay, Richardson, Spence and Brand2000; Thrush & Dayton, Reference Thrush and Dayton2002; Gray et al., Reference Gray, Dayton, Thrush and Kaiser2006). A number of studies have investigated the impact of trawling on various components of marine ecosystems (e.g. Hansson et al., Reference Hansson, Lindegarth, Valentinsson and Ulmestrand2000; Drabsch et al., Reference Drabsch, Tanner and Connell2001; Jennings et al., Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001; Sparks-McConkey & Watling, Reference Sparks-McConkey and Watling2001; Thrush & Dayton, Reference Thrush and Dayton2002; Nilsson & Rosenberg, Reference Nilsson and Rosenberg2003; Rosenberg et al., Reference Rosenberg, Nilssona, Gremare and Amouroux2003; Tillin et al., Reference Tillin, Hiddink, Jennings and Kaiser2006). In general, these studies have concluded that large-scale trawling leads to less biomass and poorer diversity of benthic organisms, resulting in reduced productivity. Repeated habitat disturbance, e.g. large-scale demersal trawl fishing, has been shown to lead to an abundance of small-bodied, opportunistic, short-lived (r-selected) species with a concomitant loss of larger-bodied, longer lived, slower growing (K-selected) species (Jennings et al., Reference Jennings, Alvsvag, Cotter, Ehrich, Greenstreet, Jarre-Teichmann, Mergardt, Rijnsdorp and Smedstad1999, Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001; Ball et al., Reference Ball, Fox and Munday2000; Sparks-McConkey & Watling, Reference Sparks-McConkey and Watling2001). Changes in the composition of benthic assemblages may result in changes in the ecological functioning of the system. Apart from conserving general biodiversity, functional diversity and redundancy should also be considered in management of the integrity of a biological community (Bellwood et al., Reference Bellword, Hoey and Choat2003; FAO, 2003). Indeed, the range and contribution of the functional traits of species in a community will determine its ecosystem functional diversity (Tillin et al., Reference Tillin, Hiddink, Jennings and Kaiser2006). Maintaining an ecosystem's functional diversity can provide a buffer against large environmental shifts caused by either natural effects or anthropogenic factors (Folke et al., Reference Folke, Carpenter, Walker, Scheffer, Elmqvist, Gunderson and Holling2004; Bremner, Reference Bremner2008). Although less is known about functional redundancy, it too is likely to be linked with diversity and to be an important factor in functional performance following environmental disturbances.
The roles performed by benthic species are important in regulating ecosystem processes and can be portrayed by the biological traits they exhibit (Snelgrove, Reference Snelgrove1998). Biological traits analysis (BTA) uses a series of life history, morphological and behavioural characteristics of the species present in assemblages to indicate aspects of their ecological functioning (Chevenet et al., Reference Chevenet, Doledec and Chessel1994; Doledec et al., Reference Doledec, Statzner and Bournard1999; Charvet et al., Reference Charvet, Statzner, Usseglio-Polatera and Dumont2000; Bremner et al., Reference Bremner, Rogers and Frid2006b). No single parameter can provide a complete measure of ecosystem functioning as a whole; simultaneously considering multiple variables is considered to be more effective (Bremner, Reference Bremner2008). The BTA incorporates information both on the relative biomass of species and on a wide variety of biological characteristics (trait information). Changes in the biomass (or abundance) of taxa resulting in changes in the patterns of trait expression within assemblages, may be indicative of the effects of human impacts on ecological functioning (Bremner et al., Reference Bremner, Rogers and Frid2006a) and are often reflected by an increase in those species able to withstand the impacts. Hence, BTA is a useful tool to study the effects of anthropogenic impacts (e.g. fishing) on benthic assemblage functioning, and can also be used to monitor changes in ecosystem functioning (Bremner et al., Reference Bremner, Rogers and Frid2003; Bremner, Reference Bremner2008).
Tyler et al. (Reference Tyler, Somerfield, Van den Berghe, Bremner, Jackson, Langmead, Palomares and Webb2012) have assessed the availability of biological trait data for the well-documented demersal marine fauna of the British Isles. Based on data of 973 species from 15 phyla and 40 classes from extensive surveys around the UK, they quantified the availability of data on eight basic biological traits for each species from online databases. They found full biological data for only 9% of species (mostly fish) and 20% of species lacked any trait data. This indicates the need for basic biological data on marine species in order to properly understand their roles in marine communities and ecosystems. The need for basic trait data is even more acute in the less-known waters of southern Africa.
In order to understand the long-term impacts of fishing, studies at the scale of the fishery are required (Berkes et al., Reference Berkes, Mahon, McConney, Pollnac and Pomeroy2001). However, only a few BTA studies of faunal functional traits in relation to ecological functioning and the effects of fishing have been conducted, and these have only been conducted in European or Canadian waters: temporal changes in benthos of the North Sea (Bremner et al., Reference Bremner, Frid, Rogers, Barnes and Thomas2005); epifaunal changes in demersal trawl grounds of the North Sea (Tillin et al., Reference Tillin, Hiddink, Jennings and Kaiser2006); infaunal and epifaunal changes in fished areas of the Mediterranean Sea (de Juan et al., Reference de Juan, Thrush and Demestre2007); long-term changes in epifauna in the Bay of Fundy, Canada (Kenchington et al., Reference Kenchington, Kenchington, Henry, Fuller and Gonzalez2007); and dredging recovery of macrofauna in the English Channel (Cooper et al., Reference Cooper, Barrio Froján, Defew, Curtis, Fleddum, Brooks and Paterson2008).
This study is the first application of BTA to examine impacts of demersal trawling on marine benthic assemblages in deep-water fishing grounds (348–436 m) in a major upwelling ecosystem along the west coast of southern Africa.
Atkinson et al. (Reference Atkinson, Field and Hutchings2011) analysed benthic species biomass and abundance data to assess the effects of different levels of trawling on assemblage composition, compared benthic assemblage spatial differences (over a north–south area in excess of 800 km) and the effects of sediment type. The analyses were complicated by pseudo-replication difficulties, which are addressed by Atkinson et al. (Reference Atkinson, Field and Hutchings2011). This study re-analyses some of the same data but at the level of biological traits to address the following questions:
-
1. Do biological traits of infauna and epifauna respond to trawling effects in the same way?
-
2. What traits are characteristic of heavily and lightly trawled areas?
-
3. Do more intensely fished areas result in fewer K-selected species and/or more r-selected species?
-
4. Can BTA serve as a useful indicator for monitoring changes in fished communities?
MATERIALS AND METHODS
Sampling design
A detailed description of the sampling design and procedure is described in Atkinson et al. (Reference Atkinson, Field and Hutchings2011). In summary, benthic macrofauna were collected from aboard the RV ‘Dr Fridtjof Nansen’ in April 2007 and FRS ‘Ellen Kuzwayo’ in February 2008 (Figure 1). The dates of sampling were fortuitous, depending upon ship availability. Four sites were sampled for infauna: Namibia (Nam); Childs Bank (Child); Cape Columbine (Col); and Cape Point (Point). Two sites were sampled for epifauna (Nam and Child) with areas of heavy trawling (HT) and light trawling (LT) being sampled at each site. Atkinson et al. (Reference Atkinson, Field and Hutchings2011) defined heavily trawled sites to be those trawled between 1.5 to 2.7 times per year, while lightly fished sites were trawled between 0.1 to 1.1 times per year. Atkinson et al. (Reference Atkinson, Field and Hutchings2011) provides further detail on the distinction between HT and LT areas in terms of trawl tracks, hours fished and proportion of swept area at each site sampled in this study.
Fig. 1. Location of four selected sample sites. Black circles represent heavily trawled areas and grey circles represent lightly trawled areas (Atkinson et al., Reference Atkinson, Field and Hutchings2011).
A 0.2 m2 van Veen grab was used to collect five quantitative replicate macro-benthic infaunal samples at each of the HT and LT area sites with all infauna retained after sieving with a 1 mm mesh sieve. Macro-benthic epifauna were sampled with three semi-quantitative replicate trawls with a 1 cm mesh net liner in HT and LT areas at the Namibian and Childs Bank sites. Sediment samples collected by grab at each area were analysed for particle grain size and total organic carbon (TOC) content. Details of field sampling and laboratory processing are reported in Atkinson et al. (Reference Atkinson, Field and Hutchings2011). In brief, the sediments at both HT and LT areas at Namibia, Childs Bank and Cape Point sites were classified as ‘sand’ or ‘muddy sand’, comprising 72–89% sand, whereas at Cape Columbine, both HT and LT areas were more silty, with 52% and 21% sand, respectively. The level of TOC content varied between 0.64% and 5.55% for HT and LT areas at Namibia, Cape Columbine and Cape Point sites. However, higher TOC content was noted for HT (14.23%) and LT (11.37%) areas at Childs Bank.
Biological traits analysis (BTA)
The analysis of results generally followed the method reported in Bremner et al. (Reference Bremner, Rogers and Frid2006a) and Tillin et al. (Reference Tillin, Hiddink, Jennings and Kaiser2006). A biological trait database of all species sampled in this study was developed by compiling information from a range of literature sources, e.g. scientific publications, theses, web databases, general field books, technical papers and expert knowledge. Most of the trait information came from the species level since many traits were recorded through observations and measurements, e.g. size, body form, etc. However, the feeding types were categorized by genus or family level from published literature, mostly from the northern hemisphere. In this study, eight biological traits with a total of 42 categories were identified for the infaunal species analysis (Table 1) while nine biological traits with 41 categories were selected for the epifaunal analysis (Table 2). These basic traits are very similar to those analysed by Tyler et al. (Reference Tyler, Somerfield, Van den Berghe, Bremner, Jackson, Langmead, Palomares and Webb2012), although we were unable to find enough data on fecundity and lifespan to include these traits. Each category was scored according to the affinity of each taxon for each trait category, ranging from 0–3, where 0 = no affinity and 3 = complete affinity. A taxon could be allocated several scores for the same trait, referred to as ‘fuzzy coding’ (Chevenet et al., Reference Chevenet, Doledec and Chessel1994), e.g. one species with two types of feeding strategies was given the affinity 2 in both feeding categories. In the event that no information was available for a trait at the species level, a search was conducted at the genus level, and if still no information was found, then at the family level. To investigate any differences in body size between heavily and lightly trawled sites, the average length of each infaunal taxon was measured and scored as affinities. As an indication of the body size of epifaunal species, the average biomass of taxa occurring was categorized on a log-based scale, because the trawl samples were not fully quantitative (see Atkinson et al., Reference Atkinson, Field and Hutchings2011) and scored as affinities.
Table 1. Infauna: biological traits and categories used in the analysis. Each category is allocated affinities ranging from 0–3 where 0 = no affinity and 3 = complete affinity. Size was analysed for both measured size at capture (NSA) and normal adult size (NSB) obtained from literature.
Table 2. Epifauna: biological traits and categories used in the analysis. Each category is allocated affinities ranging from 0–3 where 0 = no affinity and 3 = complete affinity. Maximum adult size (MS) was obtained from literature.
The ‘fuzzy coded’ species by traits matrices were weighted by their biomass at each site through matrix multiplication. The biomass-weighted trait category scores were summed over all taxa present at the area, providing a measure of frequency of occurrence of trait categories over the whole assemblage (Charvet et al., Reference Charvet, Statzner, Usseglio-Polatera and Dumont2000; Bremner et al., Reference Bremner, Rogers and Frid2006a). The traits weighted by biomass matrices were analysed independently for infaunal and epifaunal assemblages.
The non-parametric Mann–Whitney U-test, which ranks data on an ordinal scale and makes a comparison between two allocated groups, was used to test for significant differences in infaunal and epifaunal biological traits weighted by biomass for each trawling treatment (HT and LT). The same test was then used to detect significant differences in biological traits of infauna between low (≤72%) and high (>72%) proportions of sand and low (<20%) and high (≥20%) proportions of mud using STATISTICA v.8 software. The classification of sediment (i.e. low or high percentage sand or mud) was identical at paired HT and LT areas except at Cape Columbine where the sand content was classified as high in HT areas and low in LT areas (Atkinson et al., Reference Atkinson, Field and Hutchings2011).
Bubble plots, scaled to represent the biomass of infaunal species exhibiting significant traits at each site, were overlaid on principal coordinate analysis (PCO) plots of 4th root transformed infaunal biomass. The conservative 4th root transformation was adopted in this study to down-weight the excessive contributions of abundant species to the similarities calculated between species (Field et al. Reference Field, Clarke and Warwick1982; Clark & Gorley, 2006). Similarly, epifaunal species with significant traits were represented by vectors overlaid on a PCO plot of log-scale categorized biomass at each area. The PCO routine ordinates the data onto Euclidean axes by minimizing residual variation in the dissimilarities of the Bray–Curtis measure (Clarke & Warwick, Reference Clarke and Warwick2001; Clarke & Gorley, Reference Clarke and Gorley2006). The PCO was performed using PRIMER v.6 and its add-on package PERMANOVA+ (Clarke & Warwick, Reference Clarke and Warwick2001; Clarke & Gorley, Reference Clarke and Gorley2006; Anderson et al., Reference Anderson, Gorley and Clarke2008).
RESULTS
Infaunal BTA
A total of 248 infaunal species was identified and assigned traits scores. Seventeen per cent of infaunal biological traits tested were significantly different at the 5% level between areas of heavily and lightly trawled intensities (Table 3). Comparing the summed rank values for each significantly different infaunal trait between heavily and lightly trawled areas (calculated with the Mann–Whitney U test), it is evident that smaller (<5 mm) suspension and surface deposit feeders had a higher biomass at heavily trawled areas (Table 3). More surface crawlers with a long thin body form and high mobility occurred at lightly trawled areas. Ten and seven per cent of infaunal biological traits were significantly different at the 5% level in areas with low or high sand or mud compositions, respectively (Tables 4 and 5). Species between 5 mm and 1 cm in size, sessile, having lecithotrophic larval phases and a detritus/sandlicking feeding strategy had greater biomass in areas with more than 72% sand composition (Table 4). Surface crawlers and species having no larval life phases (direct development) had significantly greater biomass in areas with more than 20% mud composition. Sessile species had significantly less biomass in areas with high proportions of mud (Table 5). The first two axes of the infaunal biomass PCO account for only 33.6% of the total variation, suggesting a poor reflection of the structures occurring in the multivariate space in two-dimensions (Figure 2A). Nonetheless, HT and LT areas separate out at three of the four sites. An overlay of circles, scaled to represent the biomass of species with specific traits, was used to investigate the relationship of significant traits to trawling intensity. Species having small body size (<5 mm measured and 1–3 mm from literature) had greater biomass at the HT areas off Namibia and Cape Columbine (Figure 2B, C). Similarly, species that suspension-feed or deposit-feed near the sediment surface also had higher biomass at HT areas (Figure 2G, H). Species of high mobility (AM4), long thin body form (BF5) and surface crawlers (AH5) had larger biomass at the LT areas of Namibia and Childs Bank (Figure 2D–F).
Fig. 2. Infauna: principal coordinate ordination graphs reflecting 4th root transformed infaunal biomass weighted by significant biological traits from Table 1. Panel 1A shows the PCO of 40 site and area replicates using Euclidean distance on the biomass-weighted biological traits matrix. Panels 1B–11B–G show bubble plots superimposed on 1A and scaled to represent biomass distribution of each trait of each replicate sample at each site and area. PCO1, PCO2, 1st and 2nd principal coordinates; H, heavily trawled; L, lightly trawled.
Table 3. Infauna: heavily vs lightly trawled areas. Significantly different biological traits (17% of traits tested) tested by non-parametric Mann–Whitney U between heavily and lightly trawled areas for biological traits weighted by infaunal biomass. LT, lightly trawled; HT, heavily trawled. Higher rank indicates larger biomass.
Table 4. Sand content: significantly different infaunal biological traits (10% of traits tested) tested by non-parametric Mann–Whitney U between low and high sand content groups for biological traits weighted by infaunal biomass. Sand content was classified as high if >72% and low if ≤72%. Higher rank indicates larger biomass.
Table 5. Mud content: significantly different infaunal biological traits (7% of traits tested) tested by Mann–Whitney U between low and high mud content groups for biological traits weighted by infaunal biomass. Mud content was classified as high if ≥20% and low if ≤20%. Higher rank indicates larger biomass.
Epifaunal BTA
A total of 60 epifaunal species was identified to genus or species level and assigned traits scores. Twenty-four per cent of traits tested were significantly different at the 5% level between HT and LT areas (Table 6). Most (80%) of the significant traits had greater biomass in the LT areas with only traits of temporary attachment and cylindrical body form having higher biomass in the HT areas (Table 6). Species weighing less than 10 g in the samples, but able to reach a body size of 6–10 cm had greater biomass at LT areas. Similarly, a species having either medium mobility, dorsally flattened, laterally flattened or spherical body form or having feeding strategies of scraper/grazer and sub-surface deposit feeder had significantly greater biomasses at LT areas in comparison to paired HT areas (Table 6).
Table 6. Epifauna: significantly different biological traits (24% of traits tested) tested by non-parametric Mann–Whitney U between heavily and lightly trawled areas for biological traits weighted by epifaunal biomass. Higher rank indicates larger biomass.
The first two axes of the PCO plot account for 74% of the total variation indicating a good portrayal of the epifaunal biomass multivariate analysis in two dimensions (Figure 3). By superimposing a vector overlay (Pearson correlation) of significant traits onto the PCO plot, it is evident that the traits temporary attachment (DA2) and cylindrical body form (BF1) feature in species with greater biomasses at the Namibian HT area (Figure 3). The length and direction of each vector indicate the strength and sign of the relationship between that trait and the PCO axes. All other significant traits (SS1, MS3, AM3, BF2, BF3, BF4, FH2 and FH4) are as a result of species with these traits having greater biomass at either Childs Bank or Namibian LT areas.
Fig. 3. Epifauna: principal coordinate ordination plot with vector overlays scaled to represent epifaunal biomass of each significant trait distribution (see Table 2) between heavily and lightly trawled areas after log-scale categorization, based on Bray–Curtis resemblance measure. PCO1, PCO2, 1st and 2nd principal coordinates; Nam, Namibia; Child, Childs Bank.
Links between benthic communities and demersal fish
The diets of the 32 most commonly trawled demersal fish species occurring in the region (from Atkinson et al., Reference Atkinson, Field and Hutchings2011, Reference Atkinson, Jarre, Shannon, Field, Kruse, Browman, Cochrane, Evans, Jamieson, Livingston, Woodby and Zhang2012) were assessed from available literature (MacPherson et al., 1983; Russell et al., 1983; Meyer & Smale, Reference Meyer and Smale1991; Punt et al., Reference Punt, Leslie, Du Plessis, Payne, Brink, Mann and Holborn1992; Punt & Leslie, Reference Punt and Leslie1995; Pillar & Barange, Reference Pillar and Barange1997; Bianchi et al., Reference Bianchi, Carpenter, Roux, Molloy, Boyer and Boyer1999) and from information provided through FishBase (Table 7). The majority of the trawled fish community feeds mainly on pelagic species occurring in the water column (44%), 37% feed mainly on benthic species, while 19% feed on both pelagic and benthic species. Thus, although the demersal trawled fish community is associated with the benthic community, much of this association occurs indirectly through small fish species and planktonic invertebrates that occur just above the seabed or in the water column. It is probable that the trawled fish community has a stronger association with the seabed habitat than as a dominant source of prey.
Table 7. Southern Benguela demersal trawl fish diets: most abundant trawl fish species in southern Benguela fish communities taken from Atkinson et al. (Reference Atkinson, Leslie, Field and Jarre2011, Reference Atkinson, Jarre, Shannon, Field, Kruse, Browman, Cochrane, Evans, Jamieson, Livingston, Woodby and Zhang2012) with diets from literature and FishBase as indicated in table. Last column indicates classification of fish into those feeding on benthic species (fish and invertebrates occurring in, on or near the seabed) (37%), pelagic species (fish and invertebrates occurring in the water column) (44%) or both (19%).
DISCUSSION
To increase our understanding of the effects trawling may have in changing ecosystem functioning, it is essential to recognize the relationship between the biological functions of species and their vulnerability to trawl disturbance (Tillin et al., Reference Tillin, Hiddink, Jennings and Kaiser2006). The BTA is considered to be a powerful technique to evaluate aspects of the ecological functioning of benthic assemblages (Bremner et al., Reference Bremner, Rogers and Frid2006a) and was chosen in this study to compare significant biological traits in HT and LT environments. Other studies have reported on changes in some of the same traits as those investigated in this study. For example, species with smaller body sizes appear to occur more in impacted areas (Kaiser et al., Reference Kaiser, Ramsay, Richardson, Spence and Brand2000), a result similar to this study (Tables 3 and 4). Feeding type is another trait that differs between heavily and lightly trawled areas (Tables 3 and 4) and appears to significantly reflect the adaptation of the organisms to the habitat (de Juan et al., 2007). As in this study, others have also found significant increases of motile scavengers (Kaiser & Spencer, Reference Kaiser and Spencer1994; Collie et al., Reference Collie, Escanero and Valentine1997; Ramsay et al., Reference Ramsay, Kaiser and Hughes1998; Demestre et al., Reference Demestre, Sanchez, Kaiser, Kaiser and de Groot2000) and deposit feeders (Frid et al., Reference Frid, Harwood, Hall and Hall2000) in trawled areas. The movements and the organisms’ positions in the sediments were also considered important with regard to nutrient flux (Widdicombe et al., Reference Widdicombe, Austen, Olsgard, Schaanning, Dashfield and Needham2004; Olsgardet al., Reference Olsgard, Schaanning, Widdicombe, Kendall and Austen2008). Body form was found to differ between HT and LT areas for infauna (Table 3) and this trait may be useful to detect resilience to higher levels of trawling disturbance. No records were found in the literature of the significance of larval type affected by trawling but variations in latitude have been shown to influence this trait for several planktonic larvae of benthic invertebrates (Thorson, Reference Thorson1936; Schluter, Reference Schluter1998).
Infaunal versus epifaunal responses to trawling
Atkinson et al. (Reference Atkinson, Field and Hutchings2011) showed that epifaunal assemblages respond more significantly to the impacts of heavy trawling than infaunal assemblages. This result appears to manifest in particular in deeper water trawling (>200 m), since studies in shallower waters have found the opposite response (Collie et al., Reference Collie, Hall, Kaiser and Poiner2000; Jennings et al., Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001; Hinz et al., Reference Hinz, Prieto and Kaiser2009). The results from the present biological traits analysis support the findings of Atkinson et al. (Reference Atkinson, Field and Hutchings2011) with significant differences detected in a greater number of epifaunal traits (24% of traits measured) than infaunal traits (17% of traits measured) in HT and LT areas of the southern Benguela region. This indicates that an assessment of benthic assemblage biological traits is sufficiently sensitive to detect changes in benthic community function resulting from a trawling disturbance. The current study reports on the analysis of four sites for infauna and two sites for epifauna. In spite of the small sample size, the response of the biological traits suggests a likely effect on benthic community functioning, manifesting to a greater extent in epifaunal species. Significant differences in infaunal biological traits also occurred with differences in the proportions of sand (12% of biological traits) and mud (7% of biological traits). This suggests that infaunal community functioning is also influenced by the sediment properties. Schratzberger et al. (Reference Schratzberger, Warr and Rogers2007) used trait composition in meiofauna to investigate which environmental variables influenced the communities. In their study, biological traits were also found to be significantly correlated to proportions of sand and mud.
Epifaunal species having biological traits of temporary attachment and cylindrical body form occur in greater biomass in areas of more intense trawling. In this study, the burrowing anemone, Actinauge granulata, is largely responsible for these traits and indeed appears to prevail in areas subjected to heavy trawling (see Atkinson et al., Reference Atkinson, Field and Hutchings2011). All other significant biological traits analysed were represented by a greater biomass occurring in LT areas. In this study, most of the species with a dorsally flattened body form were represented by starfish and crabs, while laterally flattened body forms were represented by three prawn species known to feed on or near the seabed. Various sponge species were categorized as having a spherical body form and occurred in greater biomass in LT areas. Similarly, three urchin species (Echinus gilchristi, Brissopsis lyrifera capensis and Spatangus capensis) with feeding strategies of either scraper or subsurface deposit feeder occurred in greater biomass in the LT areas. These widely diverse species (starfish, crabs, benthic prawns, sponges and urchins) appear to be sensitive to intense levels of trawling.
Traits related to LT areas
The long, thin body form of species like polychaetes, predicted to proliferate in areas of disturbance where larger macrofauna are removed (Bergman & van Santbrink, Reference Bergman and van Sandbrink2000; Kaiser et al., Reference Kaiser, Ramsay, Richardson, Spence and Brand2000), were more abundant in the LT areas in this study (Table 3). Similarly, other studies (Jennings et al., Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001; Hinz et al., Reference Hinz, Prieto and Kaiser2009) did not show consistent increases in small polychaete species in response to trawl disturbance. Jennings et al. (Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001) reported that at moderate levels of disturbance there was some evidence for proliferation of small polychaetes, but at higher levels of disturbance, their biomass was reduced. Hinz et al. (Reference Hinz, Prieto and Kaiser2009) reported that overall small polychaetes did not respond positively to increasing trawl disturbance. Nephtys spp. and Chloeia inermis accounted for the high polychaete biomass in the LT areas in this study. Such species may be vulnerable to physical damage due to intense trawling levels whilst being able to withstand lighter levels of disturbance, as proposed by Jennings et al. (Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001). This may similarly explain the greater prevalence of surface crawlers and highly mobile species in the LT areas in this study. These traits are also largely represented by Nephtys spp. and Chloeia inermis.
Habitat modification and changes in the proportions of mud and sand are reported to occur when the seabed is frequently trawled (Steele et al., Reference Steele, Alverson, Auster, Collie, DeAlteris, Deegan, Escobar-Briones, Hall, Kruse, Pomeroy, Scanlon and Weeks2002), which, in turn, can change the suitability of habitats for the organisms. Significant differences in biological traits between high and low proportions of mud detected in this study suggest that surface crawlers and species having direct larval development are more prevalent in LT, muddier environments. Such environments would be expected to be more stable due to reduced physical disturbance, thus retaining finer sediment particles (Steele et al., Reference Steele, Alverson, Auster, Collie, DeAlteris, Deegan, Escobar-Briones, Hall, Kruse, Pomeroy, Scanlon and Weeks2002). When trawling disturbs the sediment, fine mud particles are most likely to be transported away in suspension by currents near the seabed, leaving coarser particles to settle out near the trawled areas. Surface crawlers (infauna) generally occurred in greater biomass in the LT areas of this study (Table 3), indicating their vulnerability to heavier trawl activities through exposure to the passing gear on the surface of the sediment. Species with direct larval development are also likely to be vulnerable to higher levels of disturbance as the emerging juveniles, usually small in size, are more prone to local extinctions due to failure of new recruitment.
Traits related to HT areas
Several biological traits show a positive response to higher levels of trawling disturbance. Infaunal surface deposit feeders, suspension feeders and species having a small body size (<5 mm and maximum adult size of 1–3 cm) were significantly more abundant in HT areas (Table 3), possibly attracted by the increased disturbance levels leading to increased suspended food supply (Dayton et al., Reference Dayton, Thrush, Agardy and Hofman1995; Steele et al., Reference Steele, Alverson, Auster, Collie, DeAlteris, Deegan, Escobar-Briones, Hall, Kruse, Pomeroy, Scanlon and Weeks2002) or able to readily escape the impact of the fishing gear due to their small body size. Epifaunal species with cylindrical body form and temporary attachment (i.e. the burrowing anemone Actinauge granulata) had a significantly greater biomass in heavily trawled areas (Table 4). Tillin et al. (Reference Tillin, Hiddink, Jennings and Kaiser2006) reported significantly more burrowers and scavengers in areas of high fishing impact and more filter feeders and attached fauna in less disturbed areas. de Juan et al. (Reference de Juan, Thrush and Demestre2007) found greater abundance of mobile burrowing traits in the trawled area, while the untrawled area had more surface crawlers, highly mobile (similar to results of this study) and filter-feeding and deposit feeding organisms. Kenchington et al. (Reference Kenchington, Kenchington, Henry, Fuller and Gonzalez2007) observed increases in mobile species, burrowers and scavengers with a decrease in sessile, filter feeders and species having permanent tubes (e.g. fan worms) with the onset of low intensity trawl fishing in the Bay of Fundy, Canada. Species with biological traits of mobility, burrowing life habit and scavenging feeding mode appear to respond similarly to fishing disturbance by increasing in biomass (Bremner et al., Reference Bremner, Frid, Rogers, Barnes and Thomas2005; Tillin et al., Reference Tillin, Hiddink, Jennings and Kaiser2006; de Juan et al., Reference de Juan, Thrush and Demestre2007; Kenchington et al., 2007). As discussed above, this study has some similar findings to other studies conducted in fished environments (e.g. burrowing life habit) whilst an increase in scavengers in heavily fished areas was not detected in the present work.
Trawling effects on K- and r-selected traits
In this study and as predicted, infauna with smaller body size (<5 mm), an r-selected trait, occur in significantly greater biomass in the HT areas. Other studies on benthic trawling impacts have also observed a shift from large, slow growing fauna to smaller and faster growing animals with varying levels of trawl disturbance (e.g. Kaiser et al., Reference Kaiser, Ramsay, Richardson, Spence and Brand2000; Rumohr & Kujawski, Reference Rumohr and Kujawski2000; Jennings et al., Reference Jennings, Dinmore, Duplisea, Warr and Lancaster2001; Hinz et al., Reference Hinz, Prieto and Kaiser2009). Sessile species, filter feeders and those with larger body sizes (K-selected traits) appear to be more negatively impacted by fishing disturbance. The results of this study are similar to previous studies with respect to some of the traits measured (e.g. body size and mobility), but not with respect to others for which we have little or no data (e.g. feeding strategies and life habits). Some species with K-selected traits (e.g. high mobility, surface crawler) reflect high vulnerability to the impacts of trawling, whilst others (e.g. permanent attachment, large size) do not. This suggests that a simple r–K selected classification will not necessarily provide an appropriate indication of the species’ vulnerability to trawling. The variability in results obtained from BTA further illustrates the uncertainty of true quantification of the effects of fishing with respect to functional diversity, as suggested by Bremner (Reference Bremner2008). Additionally, different studies have different baseline environmental conditions against which change is measured, necessitating site-specific studies in fished habitats with appropriate comparisons to similar unfished habitats.
Application of BTA in an approach to fisheries management
Protection of habitats and their ecological functioning is fundamental to ensuring ecological sustainability and thus a key element in applying the ecosystem approach to fisheries management (Shannon et al., Reference Shannon, Cury, Nel, van der Lingen, Leslie, Brouwer, Cockcroft and Hutchings2006; Frid et al., Reference Frid, Paramor and Scott2006, Reference Frid, Paramor, Brockington and Bremner2008). The use of BTA to describe and quantify ecological functioning is growing in popularity. However, Bremner (Reference Bremner2008) cautions that such analytical methods should not be viewed as a panacea to define ecological functioning. Whole ecosystem functioning should contain elements of physical, chemical and biological components and should include all organism groups, i.e. micro-, meio-, macro- and mega-organisms, their interactions and energy flows. As an example of interactions, the diet assessment of demersal trawl fish shows only some level of interaction with benthic invertebrates and a greater reliance on small, benthic dwelling fish for food. Changes in benthic invertebrate species as a result of trawling are unlikely to greatly affect demersal trawl fish communities in terms of prey availability. However, the interaction between demersal trawl fish and benthic habitat structure and features is likely to be stronger (Auster et al., Reference Auster, Malatesta and Donaldson1997; Steele et al., Reference Steele, Alverson, Auster, Collie, DeAlteris, Deegan, Escobar-Briones, Hall, Kruse, Pomeroy, Scanlon and Weeks2002). Such interactions should be considered when taking whole ecosystem functioning into account. The BTA only addresses a small component of whole ecosystem functioning and its limitations are acknowledged. In particular, this study underlines the lack of information on basic biological traits and life history studies of macro-benthic invertebrates in southern African waters and worldwide (Tyler et al., Reference Tyler, Somerfield, Van den Berghe, Bremner, Jackson, Langmead, Palomares and Webb2012). Nevertheless, BTA is considered to be a useful tool to distil important benthic faunal information for further development into indicators of community functioning in fished marine systems and therefore is pertinent for implementation of fisheries management based on an ecosystem approach.
ACKNOWLEDGEMENTS
We would like to acknowledge our dear friend, colleague and supervisor, the late Professor J.S. Gray (1941–2007) who initiated this project. We thank R. Warwick and P. Somerfield for discussions and help in identifying biological traits. We also thank A. Bjorgesaeter, S. Mafwila and L. Hutchings for their assistance with field sampling, and R Leslie for his advice on fish diets. Many thanks are also due to all the crew members from the RV ‘Dr Fridtjof Nansen’ and FRS ‘Ellen Kuzwayo’ for hosting us and assisting with sampling. We are grateful to two anonymous referees who made c0nstructive comments on the manuscript.
FINANCIAL SUPPORT
This study forms part of the NORSA 3004 project from which funding was received.
